Covid Data Trends and Statistics

By: Jacob Siegel

Abstract

There is a wide rage in the death rate attributed to the Corona Virus between the US states in the time frame from the onset of the Pandemic to August 2021. The range is from 0.038 % of the population in Hawaii to 0.3% of the population in New York, nearly an order of magnitude difference. To understand this range in outcomes, a state-level dataset was assemble with various attributes pertaining to each state such as average obesity, average GDP, education level, race, population density, political party of the governor, etc. This dataset was then used as the input for sklearn’s linear regression model RidgeCV, a ridge regression with build in cross-validation. This analysis shows that many factors contributed to this range in death rate with the most salient features being the population density, obesity rate, and education level of the state. The political party of the state’s leadership (Governor as of 2020) does not appear to have a strong influence on the death rate, however, news and social media have fomented an intense belief that one political party is better at managing the corona virus than another, creating a strong divide and leaving both sides attributing failure to the other.

#Import Libraries
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline 
import seaborn as sns

#Import sklearn features
from sklearn import metrics
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn import tree
from sklearn.metrics import classification_report,confusion_matrix
from sklearn.metrics import roc_auc_score
from sklearn.metrics import roc_curve
from sklearn.model_selection import GridSearchCV
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split, StratifiedKFold, cross_val_score
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
from sklearn import preprocessing
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import confusion_matrix, classification_report, accuracy_score, precision_score, recall_score, f1_score
from sklearn.preprocessing import MinMaxScaler
from sklearn.linear_model import RidgeCV, Ridge, Lasso
from sklearn.preprocessing import MinMaxScaler

#Import Statsmodel features
import statsmodels.api as sm
from scipy.stats import zscore
from statsmodels.stats.outliers_influence import variance_inflation_factor
from scipy import stats
!pip install plotly==5.3.1
import plotly.express as px 

#Stop Warmings
import warnings
warnings.filterwarnings("ignore")

Requirement already satisfied: plotly==5.3.1 in c:\winapps\anaconda\lib\site-packages (5.3.1)
Requirement already satisfied: tenacity>=6.2.0 in c:\winapps\anaconda\lib\site-packages (from plotly==5.3.1) (8.0.1)
Requirement already satisfied: six in c:\winapps\anaconda\lib\site-packages (from plotly==5.3.1) (1.15.0)

1.0 Data Overview

The data is a state-level overview of statistics obtained from a variety of sources including the CDC, Bureau of Labor Statistics, KFF and Wikipedia. The columns and source are as follows:

Columns:

• State: Name of State
• Population: Population of state
• Political Party: Governor of State during 2020
• Election: 2016 Presidential Election Results
• Cases: Total cases recorded for state as of August 1, 2021
• Deaths: Total deaths recorded for state as of August 1, 2021
• Tests: Total tests recorded for state as of August 1, 2021
• Vaccine: Total vaccines recorded for state as of August 1, 2021
• Unemployment: Average unemployment rate for state
• GDP: Average GDP per capita per state (US$)
• Area: Area of State (miles^2)
• White/Black/Hispanic/Asia: Percentage of each race in the state
• High_Shool/College: Percentage of state with a high school or college degree
• Obesity: percentage of state considered Obese
• Life Expectancy: Average life expectancy of the state
• Med_Age: Median age per state

Sources:

• https://www.worldometers.info/coronavirus/country/us/
• https://en.wikipedia.org/wiki/List_of_current_United_States_governors
• https://worldpopulationreview.com/state-rankings/educational-attainment-by-state
• https://www.bls.gov/web/laus/laumstrk.htm
• https://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_GDP_per_capita
• https://www.cdc.gov/obesity/data/prevalence-maps.html
• https://www.kff.org/other/state-indicator/distribution-by-raceethnicity/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D

#Load data set
States = pd.read_csv("States.csv")
States.sample(2)

	State	Population	Political_Party	Election	Cases	Deaths	Tests	Vaccine	Unemployment	GDP	Area	White	Black	Hispanic	Asian	High_School	College	Obesity	Life_Expectancy	Med_Age
17	Louisiana	4648794	Democratic	Republican	573903	11210	8499174	3783465	6.9	43917	52378.13	0.588	0.319	0.052	0.017	0.85	0.24	35.9	76.1	37.5
8	Florida	21477737	Republican	Republican	2770817	39695	33718946	23116303	5.0	39543	65757.70	0.530	0.150	0.266	0.027	0.88	0.30	27.0	80.2	42.5

A second data set of deaths recorded in 1 week during August to highligh recent changes and more vacinations.

#Appendix data for deatrhs in the past 7 days per state
States7 = pd.read_csv("Last_7_Days.csv")
States7.sample(2)

	State	Deaths_7_Days
14	Iowa	39
9	Georgia	470

States.describe().round(1).T

	count	mean	std	min	25%	50%	75%	max
Population	50.0	6551895.9	7389342.9	578759.0	1827712.2	4558233.5	7530849.0	39512223.0
Cases	50.0	712852.0	811884.1	25320.0	206516.5	492664.0	880914.0	4046728.0
Deaths	50.0	12285.9	14422.9	260.0	2556.2	7576.0	13974.5	64786.0
Tests	50.0	10702610.2	14409660.6	456246.0	2878415.5	5585867.5	11435898.5	75404964.0
Vaccine	50.0	6884604.4	8229252.3	452145.0	1790883.0	4388322.5	9104884.5	45890960.0
Unemployment	50.0	5.2	1.6	2.5	4.0	5.0	6.4	7.9
GDP	50.0	48399.1	8764.9	31881.0	43279.5	47416.5	53772.2	65545.0
Area	50.0	75933.5	97327.4	1544.9	37416.6	57093.2	84564.9	665384.0
White	50.0	0.7	0.2	0.2	0.6	0.7	0.8	0.9
Black	50.0	0.1	0.1	0.0	0.0	0.1	0.1	0.4
Hispanic	50.0	0.1	0.1	0.0	0.1	0.1	0.1	0.5
Asian	50.0	0.0	0.1	0.0	0.0	0.0	0.0	0.4
High_School	50.0	0.9	0.0	0.8	0.9	0.9	0.9	0.9
College	50.0	0.3	0.1	0.2	0.3	0.3	0.3	0.4
Obesity	50.0	32.1	3.9	23.8	29.3	32.1	34.8	40.8
Life_Expectancy	50.0	78.8	1.8	74.8	77.9	79.1	79.9	82.3
Med_Age	50.0	38.8	2.3	31.3	37.6	38.8	39.9	45.0

1.1 Feature Engineering

Case_Ave = States.Cases.sum() / States.Population.sum()   #Percentage of Cases for all US population
Death_Ave = States.Deaths.sum() / States.Population.sum() #Percentage of Deaths for all US Popilatiion
MorRate = States.Deaths.sum() / States.Cases.sum()        #mortality rate in the US
print('Percentage of cases for all US population:', round(Case_Ave*100, 2),'%')
print('The US mortality rate is:', round(MorRate*100, 2), '%')
print('Total deaths in the US is:', States.Deaths.sum())

Percentage of cases for all US population: 10.88 %
The US mortality rate is: 1.72 %
Total deaths in the US is: 614295

#Marker for state above or below national case average (to be used for logistics regression): 
States['Case_Compare'] = (States.Cases / States.Population > Case_Ave)*1 

#Marker for state above or below national death average (to be used for logistics regression):
States['Death_Compare'] = (States.Deaths / States.Population > Death_Ave)*1

#Positivity Rate of the tests (Toral Cases / Total Tests):
States['Percent_Tests'] = (States.Cases / States.Tests) *100

#Percentage of the state vacinated:
States['Percent_Vaccine'] = (States.Vaccine / States.Population) * 100

#Percentage of the state that has died (Total Deaths / Total Polulation):
States['Percent_Death'] = (States.Deaths / States.Population) * 100

#Percentage of the state that has died in the past 7 days (Total Deaths / Total Polulation):
States['Percent_Death_7'] = (States7.Deaths_7_Days / States.Population) * 100

#Percentage of the state that has died (Total Deaths / Total Polulation):
States['Percent_Cases'] = (States.Cases / States.Population) * 100

#Population Density of the state (Polpulation / Area):
States['Pop_Density'] = (States.Population / States.Area) 

#Marker for political party
States['Political_Party_N'] = States.Political_Party.apply(lambda x: '0' if x == 'Republican' else 1)
States['Political_Party_N'] = States['Political_Party_N'].astype('int64')

#Marker for Election
States['Election_N'] = States.Election.apply(lambda x: '0' if x == 'Republican' else 1)
States['Election_N'] = States['Election_N'].astype('int64')

print('Average death rate for all US population:', round(Death_Ave*100, 2),'%')
print('Minimum death rate from a state:', round(States['Percent_Death'].min(),3))
print('Maximum death rate from a state:', round(States['Percent_Death'].max(),3))
print('Standard deviation of death rate from all states:', round(States['Percent_Death'].std(),3))

plt.figure(figsize=(8,3))
plt.suptitle('Distribution of Death Rate From Every State', fontsize=16)
plt.subplot(1,2,1)
sns.distplot(States['Percent_Death'], color="grey")
plt.subplot(1,2,2)
sns.boxplot(x='Percent_Death',data=States, color="grey")

Average death rate for all US population: 0.19 %
Minimum death rate from a state: 0.038
Maximum death rate from a state: 0.3
Standard deviation of death rate from all states: 0.063

<AxesSubplot:xlabel='Percent_Death'>

2.0 EDA: Univariate Analysis With a Focus on Political Party

plt.figure(figsize=(13,5))
plt.suptitle('Cases and Deaths vs. Polulation By Political Party', fontsize=16)

j=1
for i in ['Cases', 'Deaths']:
    plt.subplot(1,2,j)
    sns.regplot(x="Population", y=i, data = States[States['Political_Party'] != 'Republican'], 
                color="b", label='Democrat')
    sns.regplot(x="Population", y=i, data = States[States['Political_Party'] != 'Democratic'], 
                color="r", label='Republican')
    plt.yscale("log")
    plt.xscale("log")
    plt.legend()  
    j=j+1

Observation:

• The trend of state population vs Cases and Deaths in not surprising, and some scatter can be seen in the data. The trend for Deaths vs Population is nearly identical for both political parties, while the cases trend are slightly higher in Republic states.

plt.figure(figsize=(12,4))
plt.suptitle('Average Cases, Deaths, Vaccines, Unemployment and GDP By Political Party', fontsize=16)
j=1
for i in ['Cases', 'Deaths', 'Vaccine']:
    plt.subplot(1,5,j)
    x = ['Republican', 'Democrat']
    y = np.array([(States[States['Political_Party'] != 'Democratic'][i].sum() / 
                   States[States['Political_Party'] != 'Democratic'].Population.sum()) * 100, 
                  (States[States['Political_Party'] != 'Republican'][i].sum() / 
                   States[States['Political_Party'] != 'Republican'].Population.sum()) * 100])
    plt.bar(x, y, color=['red', 'blue'])
    plt.tight_layout()
    plt.title(i)
    j=j+1

#------------------------------------------------------------------------------

for i in ['Unemployment', 'GDP']:
    x = ['Republican', 'Democrat']
    y = np.array([(np.array([States[States['Political_Party'] != 'Democratic'].Population * 
                             States[States['Political_Party'] != 'Democratic'][i]]).sum()) / 
                  States[States['Political_Party'] != 'Democratic'].Population.sum(),
                  (np.array([States[States['Political_Party'] != 'Republican'].Population * 
                             States[States['Political_Party'] != 'Republican'][i]]).sum()) / 
                  States[States['Political_Party'] != 'Republican'].Population.sum()])
    plt.subplot(1,5,j)
    plt.bar(x, y, color=['red', 'blue'])
    plt.tight_layout()
    plt.title(i)
    j=j+1

Observation:

• As seen from the previous trend, Republican states have a higher average number of cases while the average death rate for republican and democratic states is nearly the same.
• Democratic states have a higher percentage of Vaccinations. The bar exceeds 100 percent as data is total number of shots for the states despite the fact that most people receive 2.
• It is interesting to note that Republican states have lower average unemployment with democratic states have higher average GDP, thus both groups exceed in an important measure of the economy.

plt.figure(figsize = (10,8)) # To resize the plot
plt.suptitle('Distribution of States Percentage Death (Entire Year) and Recent (7 Day) Percent Death By Political Party', fontsize=16)
j=1
for i in ['Percent_Death', 'Percent_Death_7']:
    plt.subplot(1,2,j)
    States['temp'] = States.Political_Party_N * States[i]
    sns.barplot(x=i, y="State", data=States.sort_values(i, ascending=False), color="r")
    sns.barplot(x="temp", y="State", data=States.sort_values(i, ascending=False), color="b") 
    plt.title(i, fontsize=16)
    plt.xlabel(' ')
    plt.tight_layout()
    j=j+1

Observations:

• Over the full Pandemic there was an even distribution of Republican and Democratic states, however, the recent 7 day percent of deaths leans more Republican, likely due to the difference in vaccination rates.

2.1 Distribution of variables

Calculate the variabes with the biggest difference (T Stat) do display

Rep = States[States['Political_Party'] == 'Republican'].head(23)
Dem = States[States['Political_Party'] != 'Republican']

Col = ['Unemployment', 'GDP', 'High_School', 'College', 'Obesity', 'White', 'Black', 'Hispanic', 'Asian', 
          'Percent_Vaccine', 'Life_Expectancy', 'Pop_Density', 'Med_Age']

T_Stat = []

for i in Col:
     T_Stat.append(stats.ttest_rel(Rep[i].values, Dem[i].values)[0])

T_Val = pd.DataFrame(Col ,columns =['Col'])
T_Val['T_Stat'] = T_Stat
T_Val.sort_values('T_Stat')

	Col	T_Stat
9	Percent_Vaccine	-4.351705
0	Unemployment	-3.364929
10	Life_Expectancy	-2.276905
8	Asian	-2.132513
3	College	-2.006162
12	Med_Age	-1.775448
7	Hispanic	-1.408989
1	GDP	-1.407811
11	Pop_Density	-1.209721
2	High_School	-0.049847
6	Black	0.690736
5	White	1.179617
4	Obesity	2.347456

# List of the 8 variables with the lowest T stat score (biggest difference)
toplot = T_Val.sort_values('T_Stat').iloc[0:8,0].values

plt.figure(figsize=(12,6))
plt.suptitle('Distribution Variables by Political Party', fontsize=16)
j=1
for i in toplot:
    plt.subplot(2,4,j)
    sns.boxplot(y=i, x='Political_Party', data=States, palette="seismic_r")
    plt.tight_layout()
    j=j+1

plt.figure(figsize=(12,6))
plt.suptitle('Distribution of Variables by Political Party', fontsize=16)

j=1
for i in toplot:
    plt.subplot(2,4,j)
    sns.distplot(States[States['Political_Party'] != 'Republican'][i], hist=False, color="b")
    sns.distplot(States[States['Political_Party'] != 'Democratic'][i], hist=False, color="r")
    plt.tight_layout()
    j=j+1

3.0 EDA: Multi Variate

States_Sub = States[['Percent_Death','Unemployment', 'GDP', 'High_School', 'College', 'Obesity', 
                     'White', 'Black', 'Hispanic', 'Asian', 'Percent_Vaccine', 'Life_Expectancy',
                     'Political_Party_N', 'Pop_Density', 'Med_Age', 'Election_N']]

States_Cor = pd.DataFrame(States_Sub.corr().Percent_Death.abs().sort_values(ascending=False).reset_index().drop(0,axis=0))
States_Cor.columns = ['Category', 'Correlation']
print('table of correlation values with percentage of death')
States_Cor

table of correlation values with percentage of death

	Category	Correlation
1	High_School	0.489298
2	Pop_Density	0.480644
3	Black	0.445699
4	Unemployment	0.287978
5	Obesity	0.280067
6	Life_Expectancy	0.247336
7	Asian	0.231044
8	Hispanic	0.202647
9	White	0.197065
10	Percent_Vaccine	0.170903
11	College	0.147099
12	Election_N	0.070120
13	GDP	0.013277
14	Political_Party_N	0.004680
15	Med_Age	0.002456

plt.figure(figsize=(12,11))
plt.suptitle('Corrrelations with Death Rate', fontsize=16)

j=1
for i in States_Cor.Category[0:16]:
    plt.subplot(4,4,j)
    plt.title(round(States_Cor.Correlation[j], 2), fontsize=10)
    sns.regplot(x=i, y='Percent_Death', data=States, color="black")
    plt.tight_layout()
    j=j+1

There is an outlier in the Asian column

States[States['Asian'] > .3]

	State	Population	Political_Party	Election	Cases	Deaths	Tests	Vaccine	Unemployment	GDP	...	Death_Compare	Percent_Tests	Percent_Vaccine	Percent_Death	Percent_Death_7	Percent_Cases	Pop_Density	Political_Party_N	Election_N	temp
10	Hawaii	1415872	Democratic	Democrat	46503	542	1908582	1731970	7.7	51277	...	0	2.436521	122.325323	0.03828	0.001978	3.284407	129.5196	1	1	0.001978

1 rows × 31 columns

Remove Hawaii as an outlier for asian population

States.loc[States['Asian'] > .3, 'Asian'] = States.Asian.mean()

sns.regplot(x='Asian', y='Percent_Death', data=States, color="black")

<AxesSubplot:xlabel='Asian', ylabel='Percent_Death'>

#Correlation plot
temp = States[['Unemployment', 'GDP', 'High_School', 'College', 'Obesity', 'Black', 'Hispanic', 'Asian',
            'Percent_Vaccine', 'Life_Expectancy', 'Pop_Density', 'Med_Age']]
plt.figure(figsize=(7,7))
sns.heatmap(temp.corr(), annot=True, linewidths=.5, fmt= '.1f', center = 0, cmap="YlGnBu" )  # heatmap
plt.show()

plt.figure(figsize=(12,4))

plt.subplot(1,3,1)
sns.regplot(x='College', y='Percent_Vaccine', data=States, color="black")

plt.subplot(1,3,2)
sns.regplot(x='College', y='Life_Expectancy', data=States, color="black")

plt.subplot(1,3,3)
sns.regplot(x='Obesity', y='Life_Expectancy', data=States, color="black")

<AxesSubplot:xlabel='Obesity', ylabel='Life_Expectancy'>

temp['Political_Party'] = States['Political_Party']
sns.pairplot(temp, hue = 'Political_Party', palette="seismic_r", corner=True)

<seaborn.axisgrid.PairGrid at 0x15660482700>

4.0 Linear Regression Modeling

4.1 Create Dependent (Y) and Independendt (X) variables

Create dependent(y) and independent (x) variables. X is then scaled per column.

X = States[['Unemployment', 'GDP', 'High_School', 'College', 'Obesity', 'Black', 'Hispanic', 'Asian',
            'Percent_Vaccine', 'Life_Expectancy', 'Political_Party_N', 'Pop_Density', 'Election_N', 'Med_Age']]
Y = States[['Percent_Death']]
X.iloc[:,:] = MinMaxScaler().fit_transform(X) #scale data

Check Multicollinearity using VIF scores

vif_series1 = pd.Series([variance_inflation_factor(X.values,i) for i in range(X.shape[1])],index=X.columns)
pd.DataFrame(vif_series1, columns={'VIF'}).sort_values(['VIF'], ascending=False)

	VIF
College	45.304205
Life_Expectancy	43.161759
High_School	39.596605
Percent_Vaccine	25.718074
Med_Age	23.492489
Obesity	15.489973
GDP	15.297951
Unemployment	13.281068
Hispanic	8.142943
Asian	7.681231
Black	6.740213
Election_N	6.084510
Pop_Density	3.780095
Political_Party_N	3.303359

College and Life_Expepectancy have the highest VIF scores and will be removed. Several atributes still have scores that are considered high (above 10),

X = X.drop(['College', 'Life_Expectancy'], axis=1)

vif_series1 = pd.Series([variance_inflation_factor(X.values,i) for i in range(X.shape[1])],index=X.columns)
pd.DataFrame(vif_series1, columns={'VIF'}).sort_values(['VIF'], ascending=False)

	VIF
Med_Age	22.702888
Percent_Vaccine	19.115201
GDP	13.947007
High_School	13.742428
Obesity	10.626273
Unemployment	10.238225
Asian	6.018169
Election_N	5.640821
Hispanic	4.203196
Black	4.172649
Pop_Density	3.381721
Political_Party_N	3.271213

Drop Med_Age

X = X.drop(['Med_Age'], axis=1)

vif_series1 = pd.Series([variance_inflation_factor(X.values,i) for i in range(X.shape[1])],index=X.columns)
pd.DataFrame(vif_series1, columns={'VIF'}).sort_values(['VIF'], ascending=False)

	VIF
Percent_Vaccine	14.323056
GDP	11.133157
High_School	10.906314
Unemployment	8.971839
Obesity	7.661214
Asian	6.018102
Election_N	5.632715
Black	4.148238
Hispanic	4.094081
Pop_Density	3.261507
Political_Party_N	3.229111

TEMP = pd.DataFrame() #Data frame to store iterative model outputs
for i in range(1,5001):
    model_N = "Model%d"%i
    X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.20)
    model = RidgeCV(alphas=[x/10 for x in range(1, 101)])
    model.fit(X_train, y_train)
    pred = model.predict(X_test)
    temp = pd.DataFrame([[model_N, r2_score(y_test, pred), model.intercept_[0]]],
                        columns=['model', 'R_Score', 'intercept'])
    temp2 = pd.DataFrame(model.coef_, columns = X_train.columns)
    TEMP = TEMP.append(temp.join(temp2))
    
best_model = TEMP.sort_values('R_Score', ascending=False).head(1) 
best_model_score = best_model.R_Score
best_model_int = best_model.intercept
best_model = best_model.drop({'model', 'R_Score', 'intercept'}, axis=1).T
best_model_values = best_model.reset_index()
best_model_values.columns = ['Category', 'Coef']
best_model_values = best_model_values.sort_values('Coef', ascending=False)
print('R2 score for the best model is:', best_model_score)

R2 score for the best model is: 0    0.817182
Name: R_Score, dtype: float64

#Plot of the caeficient values for linear regression
plt.title('Coeficient Values for Best Model', fontsize=16)
sns.barplot(x="Coef", y="Category", data=best_model_values, color="grey")

<AxesSubplot:title={'center':'Coeficient Values for Best Model'}, xlabel='Coef', ylabel='Category'>

best_model_predict = (X * (best_model.T.append([best_model.T]*49,ignore_index=True))).sum(axis = 1) + ((np.ones(50))*best_model_int[0])
plt.figure(figsize=(5,5))
x = [min(best_model_predict)*.95, max(best_model_predict)*1.05]
y = [min(best_model_predict)*.95, max(best_model_predict)*1.05]
plt.plot(x, y, '-r')
plt.scatter(Y, best_model_predict, color="black")
plt.title('Predicted Vs Measured Death Rate', fontsize=16)
plt.xlabel('Predicted (%)', color='black')
plt.ylabel('Actual (%)', color='black')

Text(0, 0.5, 'Actual (%)')

Rerun the model with the Y variable of percentage deaths in last 7 days

Y = States[['Percent_Death_7']]

TEMP = pd.DataFrame() #Data frame to store iterative model outputs
for i in range(1,5001):
    model_N = "Model%d"%i
    X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.20)
    model = RidgeCV(alphas=[x/10 for x in range(1, 101)])
    model.fit(X_train, y_train)
    pred = model.predict(X_test)
    temp = pd.DataFrame([[model_N, r2_score(y_test, pred), model.intercept_[0]]],
                        columns=['model', 'R_Score', 'intercept'])
    temp2 = pd.DataFrame(model.coef_, columns = X_train.columns)
    TEMP = TEMP.append(temp.join(temp2))
    
best_model = TEMP.sort_values('R_Score', ascending=False).head(1) 
best_model_score = best_model.R_Score
best_model_int = best_model.intercept
best_model = best_model.drop({'model', 'R_Score', 'intercept'}, axis=1).T
best_model_values = best_model.reset_index()
best_model_values.columns = ['Category', 'Coef']
best_model_values = best_model_values.sort_values('Coef', ascending=False)
print('R2 score for the best model is:', best_model_score)

R2 score for the best model is: 0    0.896565
Name: R_Score, dtype: float64

#Plot of the caeficient values for linear regression
plt.suptitle('Coeficient Values for Best Model', fontsize=16)
sns.barplot(x="Coef", y="Category", data=best_model_values, color="grey")

<AxesSubplot:xlabel='Coef', ylabel='Category'>

best_model_predict = (X * (best_model.T.append([best_model.T]*49,ignore_index=True))).sum(axis = 1) + ((np.ones(50))*best_model_int[0])
plt.figure(figsize=(5,5))
x = [min(best_model_predict)*.95, max(best_model_predict)*1.05]
y = [min(best_model_predict)*.95, max(best_model_predict)*1.05]
plt.plot(x, y, '-r')
plt.scatter(Y, best_model_predict, color="black")
plt.title('Predicted Vs Measured Death Rate', fontsize=16)
plt.xlabel('Predicted (%)', color='black')
plt.ylabel('Actual (%)', color='black')

Text(0, 0.5, 'Actual (%)')

5 Logistics Regression Model

Y = States[['Death_Compare']]

Y.value_counts()

Death_Compare
0                26
1                24
dtype: int64

A fairly even split of states that are above (1) and below (0) the national average

TEMP = pd.DataFrame()
for i in range(1,5001):
    model = "Model%d"%i
    X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.20)
    rf_estimator = RandomForestClassifier(random_state=1, n_estimators=10, max_depth=4)
    rf_estimator.fit(X_train,y_train)
    temp = pd.DataFrame(rf_estimator.feature_importances_, columns = [model], index = X_train.columns)
    TEMP[model] = temp[model]
    
TEMP = TEMP.sum(axis=1)/i
TEMP = TEMP.sort_values(ascending = False)
TEMP = TEMP.to_frame()
TEMP = TEMP.reset_index()
TEMP.columns = ['Category', 'Average_Importance']

plt.figure(figsize=(3,4))
plt.suptitle('Average Feature Importance From All Models', fontsize=16)
sns.barplot(x="Average_Importance", y="Category", data=TEMP, color="grey")

<AxesSubplot:xlabel='Average_Importance', ylabel='Category'>

us_state_abbrev = {
    'Alabama': 'AL',
    'Alaska': 'AK',
    'American Samoa': 'AS',
    'Arizona': 'AZ',
    'Arkansas': 'AR',
    'California': 'CA',
    'Colorado': 'CO',
    'Connecticut': 'CT',
    'Delaware': 'DE',
    'District of Columbia': 'DC',
    'Florida': 'FL',
    'Georgia': 'GA',
    'Guam': 'GU',
    'Hawaii': 'HI',
    'Idaho': 'ID',
    'Illinois': 'IL',
    'Indiana': 'IN',
    'Iowa': 'IA',
    'Kansas': 'KS',
    'Kentucky': 'KY',
    'Louisiana': 'LA',
    'Maine': 'ME',
    'Maryland': 'MD',
    'Massachusetts': 'MA',
    'Michigan': 'MI',
    'Minnesota': 'MN',
    'Mississippi': 'MS',
    'Missouri': 'MO',
    'Montana': 'MT',
    'Nebraska': 'NE',
    'Nevada': 'NV',
    'New Hampshire': 'NH',
    'New Jersey': 'NJ',
    'New Mexico': 'NM',
    'New York': 'NY',
    'North Carolina': 'NC',
    'North Dakota': 'ND',
    'Northern Mariana Islands':'MP',
    'Ohio': 'OH',
    'Oklahoma': 'OK',
    'Oregon': 'OR',
    'Pennsylvania': 'PA',
    'Puerto Rico': 'PR',
    'Rhode Island': 'RI',
    'South Carolina': 'SC',
    'South Dakota': 'SD',
    'Tennessee': 'TN',
    'Texas': 'TX',
    'Utah': 'UT',
    'Vermont': 'VT',
    'Virgin Islands': 'VI',
    'Virginia': 'VA',
    'Washington': 'WA',
    'West Virginia': 'WV',
    'Wisconsin': 'WI',
    'Wyoming': 'WY'
}

States['Abr'] = States['State'].map(us_state_abbrev) 

fig = px.choropleth(States,  # Input Pandas DataFrame
                    locations="Abr",  # DataFrame column with locations
                    color="Percent_Death_7",  # DataFrame column with color values
                    hover_name="State", # DataFrame column hover info
                    color_continuous_scale="Bluered",
                    locationmode = 'USA-states') # Set to plot as US States
fig.update_layout(
    title_text = 'State Rankings', # Create a Title
    geo_scope='usa',  # Plot only the USA instead of globe
)
fig.show()  # Output the plot to the screen